Organized by the International Working Group on Educational Data Mining.


Université du Québec à Montréal (UQAM), Canada

Machine Learning Department at the School of Computer Science, Carnegie Melon University

David E. Pritchard: Assessing Learning

Cecil and Ida Green Professor of Physics, Department of Physics, Massachustets Institute of Technology (Cambridge, MA)

Consultant, Pearson Education

Personal homepage


Historically, assessment was primarily summative – to determine the level of proficiency of an applicant, student, or a class.  To guide individual instruction in an intelligent tutoring system, or to guide a teacher whose class uses an ITS for homework, it is desirable to mine the data generated within the ITS to provide an up to date summary of each student's skill – hopefully as detailed as possible.  Such data mining is helpful for improving classroom instruction also. We'll show examples that address: "Are students learning?"  "If so, from which learning activity?" "Which parts of the tutoring work best?"  "What student habits are detrimental/helpful to learning?" and "Extending Item Response Theory to provide a Current Skill." These studies were based on data collected as students worked through problems in the online web-based tutorial program, MasteringPhysics that was used by ~100,000 college students in this past academic year.  We will also discuss with the audience, "What should we be teaching?"